Episode #78: AI Security and the New Rules of Access

Most organizations are adopting AI tools without fully reckoning with what that adoption means for security. In this episode, Kyle Hall sits down with Kyle Gotzman to break down why traditional security models, including role-based access controls and standard detection methods, weren't designed for a world where AI agents can act on behalf of users at machine speed. They cover the concept of bounded agency, why the MCP layer is becoming the critical choke point in agentic AI environments, and what IT leaders actually need to be thinking about before they hand an AI system the keys to their infrastructure. Whether your organization is deploying Copilot across the company or starting to build agentic workflows, this conversation cuts through the noise and gets practical about where the real risks live.